Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 843. (Read 2347659 times)

The quark hashrate on Nicehash/westhash has dropped from 100GHASH to 20GHASH because of private kernals.

The quark opensource kernal does 2MHASH on the AMD r9 280x
The Sharkcoin and the clones hasharate  is currently around  100 000 AMD radeon r9 280x  shitcards.
The current Etherum hashrate is 15000 AMD r9 280x cards. And they will go POS.

Out of bugdet

Neoscrypt is fucking slow with Cuda 7.5

Thanks

This is real opensource optimalizations of the most profitable algos.

Djm34 never opensource anything profitable. He just see his algos are tanking because of people like me. He used to sell his optimized lyra2v2 kernals for 1 980ti a piece. Now his work is a waste. I sold my modded private kernal to...

I just had a chance to test the new version using cuda 7.5. It is indeed a bit faster on quark.
About neoscrypt, it is unexpectedly just some percent slower than the old version, not the total mess it was when I worked on it some months ago.
That's very good news because it is much more likely to be possible to reach "full speed" or even further.
More later when I can code a bit (I already know what to try).

That is just a hardcoded string and does not reflect how ccminer was actually compiled. A more appropriate comment
would be "optimized for cuda 7.5" unless a system variable is available to read the actual compile environment.

Testing my new wattmeter, trying to break 1000W barrier with mining quark...

.74 ~87500 MHASH
.76 ~89300 MHASH

No difference in power consumption between these builds. 970-990 Watts with insane power settings. 5 x 970 + 4690K OC + 4 x 4GB, not a pure mining rig.

Oh, and my 0.05226518 BTC should arrive soon!

-edit-

standard clocks and power limit:

.74 ~76600MHASH
.76 ~79500MHASH

860-880 Watts

Way to go! I don't think anyone has managed to use cuda 7.5 to bennifit when mining until now!! on my 980ti cards 1.76 seems to be far more stable then 75, temps are cooler and big improvements on hash rates especially quark, I'm getting a stable 62mh. X13 is about 1.5MH slower then 1.74 and x11 is about 2mh slower. I will run 1.76 on a multi algo pool for the day and see how she goes with extended use.

Also should update the code with the cuda version number so it says cuda 7.5 when launched not important though lol

OK donated 0.01234BTC

Not even profiting atm, but ill do that every now and then for sure.  It's the least we can do

More than fair, I'll donate .01 now

DOGFIGHT--

If you submitted some code improvements, I'd be glad to donate.  The main reason that I have not switched immediately to the release dot 75 (and later) is the poor neoscrypt performance.

--scryptr

The Myth of the 5% faster kernel continues...   (infinite speed increase for infinite donation  )

SENT--

Still using release dot 74, here.  It seems very stable, as my "problem" rig has been mining Lyra2v2 for days without being difficult.  I'll try the new versions soon, but with altcoins not tracking up with BitCoin, I am looking for stability for the moment.

Thanks for your work, sp_ !  You'll get my continued support.       --scryptr

Guys, I have published 5% faster quark kernal for free and only recieved 0.05BTC in donations.

Do you want a continued opensource developement and a faster hashrate?
Or do I have to sell private kernals?
To keep the business profitable you need to support the developent.
0.01BTC from each of you would be nice.

Thanks very much sp_!
I'm sad as I currently am unable to test or work on it, but I hope to be able to give it a chance during the weekend.

Just tested quark release 76 on my 980ti g1 gigabyte windforce oc.
27.3 (+1,4MHASH) on the standard clocks(New WORLD RECORD)!
Release 74 is only doing 25.9MHASH.

On the EVGA SSC gtx 960. 11 MHASH on standard clocks. up 0.5 MHASH from release 74

ok great ...

then thats the first thing i will test when i get into the office tomorrow ...

a miner will hash quark on your donation address ( donate-sp.granitecoin.com ) and see how it tests ...

look out for the test on nicehash ( eu stratum ) ...

tanx again mate ... i hope it compiles and works well on the new system ...

#crysx

I just released 76 with increased hashrate all over the place. But some of the algos are still slower than release 74. Quark has the most boost this time, and mostly on compute 5.2 devices.

More cuda 7.5 modding and recoding

-Faster quark
-Faster qubit
-Faster x11
-Faster x13
-Faster x14
-Faster x15


1.5.76(sp-MOD) is available here: (11-dec-2015)

https://github.com/sp-hash/ccminer/releases/

The sourcecode is available here:

https://github.com/sp-hash/ccminer

i have mate - and i can tell you - just from the test machine its a nice chunk of hash increase ...

across thefarm - it should do nice ...

but that extra teaser for quark ... damn - if i wasnt sick - i would have already been in the office ... all of this stuff ( including the wallets and mining ) is under lock and key since the last ddos killed two of my servers and one of my vps almost got hacked ( about 16000 attempts on cracking the password in the last 48 hours ) ... so i maintain a tight ship now - and have to physically be there to get a lot of the things done ...

but i said i would put a miner to hash for you - and i will ... should be at least a beer or two in that ...

cuda 7.5 was working fine in fedora 22 x64 - as discussed previously - but im upgrading the test machine ( only one of two machines off the network - the other machine is an amd one for wolf0 ) and will try and get a compile and see how it runs when it finishes a fedora 23 x64 c7.5 upgrade ...

this nvidia test machine has 3 x gigabyte 750ti oc lp cards in it - stock ... so will luv to see more that 2800kh x11 and 5890kh quark on those cards ...

25min tests should show how they go after the upgrade - but will have to get that done tomorrow now ...

sp - still mulling over the eth kernel? ... or making it an inclusion in to ccminer-spmod76? ... ...

ive just tried following the compile directions and compile dies everywhere - mainly for llvm version conflicts in fedora 22 x64 ... but there could be much more issues ... apparently tested on fedora 23 x64 - i would like to try that when the fedora 23 x64 upgrade finishes ...

first to test with the new fedora 23 x64 c7.5 - ccminer-spmod75 tomorrow ...

tanx for the update sp ... cant wait to compile and see ...

#crysx